Asymptotic formula for the sum of a prime and a square-full number in short intervals shorter than X 1 / 2 X^{1/2}
Fumi Ogihara · 2025
Read Asymptotic formula for the sum of a prime and a square-full number in short intervals shorter than X 1 / 2 X^{1/2} by Fumi Ogihara (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-03447.txt.
Introduction
Fumi Ogihara stretches a sum law for prime-plus-rare-square counts to thin spans once past reach. The count weighs each split of a count into a prime and a square-full part, where each prime rides a squared or higher block. Past work with a co-author set the law for spans past the square-root bar. Her 2025 study proves the same law for spans that dip far short of that bar, to a set low roof built from a square-root term.
The gain widens where the law can be run: short spans are the hard yard in count craft, where sums turn wild. Prime split laws feed ciphers and pure count lore both. Each span shaved off the floor is a step in a long craft push.
Catalog record
| CREATOR | Fumi Ogihara |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-03447
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Asymptotic formula for the sum of a prime and a square-full number in short intervals shorter than X 1 / 2 X^{1/2} Fumi Ogihara Abstract . Let R ( N ) R(N) be the number of representations of N N as a sum of a prime and a square-full number weighted with logarithmic function. In 2024 2024 , the author and Y. Suzuki obtained an asymptotic formula for the sum of R ( N ) R(N) over positive integers N N in a short interval ( X X , X + H X+H ] for X 1 2 + ε ≤ H < X 1 − ε X^{\frac{1}{2}+\varepsilon}\leq H<X^{1-\varepsilon} . In this article, we improve the range of H H , that is, we prove the same asymptotic formula for X 32 − 4 15 49 + ε ≤ H ≤ X 1 − ε X^{\frac{32-4\sqrt{15}}{49}+\varepsilon}\leq H\leq X^{1-\varepsilon} . 1. Introduction The following is one of the well-known Hardy–Littlewood conjectures. Conjecture 1.1 (Hardy–Littlewood [ 2 ] ) . Every sufficiently large positive integer N is a square or the sum of a prime and a square. In other words, N=n^{2}\quad\mbox{or}\quad p+n^{2}, where p denotes a prime number and n denotes a positive integer. To approach to this conjecture, we consider the following function: R_{\textup{HL}}(N)\coloneqq\sum_{p+n^{2}=N}\log p. We call such a function a representation function. In [ 2 ] , G. H. Hardy–J. E. Littlewood also conjectured the asymptotic formula: R_{\textup{HL}}(N)\sim\mathfrak{S}(N)\sqrt{N},\quad\textup{($N$: not square)}, where \mathfrak{S}(N)\coloneqq\prod_{p>2}\bigg(1-\frac{(N/p)}{p-1}\bigg),\quad(N/p)\textup{: Legendre symbol}. In 2017, A. Languasco and A. Zaccanini considered the short interval average of the representation function: \sum_{X<N\leq X+H}R_{\textup{HL}}(N) for real numbers X , H with 4\leq H\leq X . Let C\coloneqq\exp\bigg(-c\bigg(\frac{\log X}{\log\log X}\bigg)^{\frac{1}{3}}\bigg), where c is some small positive constant which may depend on \varepsilon . Theorem 1.2 (Languasco–Zaccagnini [ 6 ] ) . For real numbers X , H with 4\leq H\leq X and any real number \varepsilon>0 , we have (1) \sum_{X<N\leq X+H}R_{\textup{HL}}(N)=HX^{\frac{1}{2}}+O(HX^{\frac{1}{2}}C^{-1}) provided X^{\frac{1}{2}}C^{-1}\leq H\leq X^{1-\varepsilon}, where the implicit constant depend only on \varepsilon . If we obtain ( 1 ) for H=1 in Theorem 1.2 , Conjecture 1.1 is true. Therefore, our goal is an improvement of the condition of H . In 2023, Y. Suzuki improved the range of H in Theorem 1.2 . Theorem 1.3 (Suzuki [ 9 ] ) . The asymptotic formula ( 1 ) provided X^{\frac{32-4\sqrt{15}}{49}+\varepsilon}\leq H\leq X^{1-\varepsilon}. In this article, we study the following function that replaces squares in R_{\textup{HL}}(N)=\sum_{p+n^{2}=N}\log p with square-full numbers: R(N)\coloneqq\sum_{\begin{subarray}{c}p+f=N\\ f\in\mathscr{Q}\end{subarray}}\log p, in which \mathscr{Q}\coloneqq\{n\in\mathbb{N}\,|\,n\mbox{ is square-full numbers}\} . Recall that a square-full number is a positive integer n such that if p is a prime dividing n , then p^{2} divides n . The author and Y. Suzuki obtained the following asymptotic formula. Theorem 1.4 ( [ 8 ] ) . For A\geq 1 , \varepsilon>0 , and X^{\frac{1}{2}+\varepsilon}\leq H\leq X^{1-\varepsilon} , we have (2) \sum_{X<N\leq X+H}R(N)=\frac{\zeta(\frac{3}{2})}{\zeta(3)}HX^{\frac{1}{2}}\Big(1+O\big((\log X)^{-A}\big)\Big), where the implicit constant depends on A,\varepsilon . Our main theorem is the following improvement of the above range of H . Theorem 1.5 . For \varepsilon>0 , we have the asymptotic formula ( 2 ) provided X^{\frac{32-4\sqrt{15}}{49}+\varepsilon}\leq H\leq X^{1-\varepsilon}. 2. Preliminary lemmas In this section, we prepare some lemmas. For a real number B\geq 2 , we define \widetilde{R}_{B}(N):=\sum_{\begin{subarray}{c}p+a^{2}b^{3}=N\\ b\leq B\end{subarray}}(\log p)\mu(b)^{2}, where \mu(n) is a Möbius function of n . Lemma 2.1 ( [ 8 ] ) . For a positive number N and a real number B\geq 1 , we have R(N)=\widetilde{R}_{B}(N)+O((N^{\frac{1}{2}}\log N)B^{-\frac{1}{2}}). By Lemma 2.1 , we have \sum_{X<N\leq X+H}R(N)=\sum_{X\leq N\leq X+H}\widetilde{R}_{B}(N)+O\Bigg(\sum_{X<N\leq X+H}(N^{\frac{1}{2}}\log N)B^{-\frac{1}{2}}\Bigg). If B=(\log X)^{4A} for A>1 , we have (3) \sum_{X<N\leq X+H}R(N)=\sum_{X\leq N\leq X+H}\widetilde{R}_{B}(N)+O(HX^{\frac{1}{2}}(\log X)^{-A}). We consider the sum of \widetilde{R}_{B}(N) on the right-hand side. Let \mathscr{Q}_{B}\coloneqq\{n\in\mathbb{N}\,|\,n=a^{2}b^{3},b\leq B\mbox{ and }b\mbox{ is square-free}\} . Then, we have \displaystyle\sum_{X<N\leq X+H}\widetilde{R}_{B}(N)= \displaystyle\sum_{\begin{subarray}{c}X<p+f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}\log p. To evaluate the short interval sum with square-full numbers, we introduce some lemmas. Lemma 2.2 ( [ 8 ] ) . Let B be a real number greater than or equal to 1 . For real numbers X , H with 4\leq H\leq X and n^{2}\leq X+H , we have \sum_{\begin{subarray}{c}n^{2}\leq f<(n+1)^{2}\\ f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}1\ll B. Lemma 2.3 . For real numbers x\leq 1 and s>1 , we have \sum_{n\leq x}\frac{\mu(n)^{2}}{n^{2}}=\frac{\zeta(s)}{\zeta(2s)}+O(x^{-s+1}), where \zeta(s) is the Riemann zeta function. Proof. We can write (4) \sum_{n\leq x}\frac{\mu(n)^{2}}{n^{2}}=\sum_{n=1}^{\infty}\frac{\mu(n)^{2}}{n^{2}}+\sum_{n>x}\frac{\mu(n)^{2}}{n^{2}}. From Euler Product Formula for the zeta function, we have \displaystyle\sum_{n=1}^{\infty}\frac{\mu(n)^{2}}{n^{2}} \displaystyle=\zeta(s)\prod_{p:prime}\bigg(1+\frac{\mu(p)^{2}}{p^{s}}\bigg)\bigg(1-\frac{1}{p^{s}}\bigg) \displaystyle=\zeta(s)\prod_{p:prime}\bigg(1-\frac{1}{p^{2s}}\bigg)=\frac{\zeta(s)}{\zeta(2s)}. The second term on the right-hand side of ( 4 ) is \sum_{n>x}\frac{\mu(n)^{2}}{n^{2}}\leq\int_{x}^{\infty}\frac{1}{u^{s}}du\ll x^{-s+1}. Substituting these formulas into ( 4 ) leads to the lemma. ∎ We first estimate the sum of the square-full numbers in the short intervals. Lemma 2.4 . Let B be a real number greater than or equal to 1 . For real numbers X , H\geq 2 , we have \sum_{\begin{subarray}{c}X<f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}1\ll HX^{-\frac{1}{2}}B+B. Proof. We write the left-hand side as \sum_{\begin{subarray}{c}X<f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}1=\sum_{X<n^{2}\leq X+H}\sum_{\begin{subarray}{c}n^{2}\leq f<(n+1)^{2}\\ f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}1. By Lemma 2.2 , we have \displaystyle\sum_{\begin{subarray}{c}X<f\leq X+H\\ f\in\mathscr{Q}_{B}\end{subarray}}1 \displaystyle\ll B\sum_{X<n^{2}\leq X+H}1 \displaystyle\leq B((X+H)^{\frac{1}{2}}-X^{\frac{1}{2}})+1 \displaystyle=B\bigg(\frac{1}{2}\int_{X}^{X+H}u^{-\frac{1}{2}}du+1\bigg)\leq HX^{-\frac{1}{2}}B+B. This completes the proof. ∎ Let S(X)\coloneqq\sum_{\begin{subarray}{c}f\leq X\\ f\in\mathscr{Q}_{B}\end{subarray}}(X-f). Lemma 2.5 . For real numbers X,H with 4\leq H\leq X , we have S(X+H)-S(X)=\frac{\zeta(\frac{3}{2})}{\zeta(3)}HX^{\frac{1}{2}}+O(HB). Proof. From the definition of \mathscr{Q}_{B} , we have \displaystyle S(X+H)-S(X) \displaystyle=H\sum_{\begin{subarray}{c}a^{2}b^{3}\leq X\\ b<B\end{subarray}}\mu(b)^{2} \displaystyle=H\sum_{b<B}\mu(b)^{2}\sum_{a^{2}\leq\frac{X}{b^{3}}}1 \displaystyle=H\bigg(X^{\frac{1}{2}}\sum_{b<B}\frac{\mu(b)^{2}}{b^{\frac{3}{2}}}+O(B)\bigg). Using Lemma 2.3 , we obtain the lemma. ∎ We second consider the Chebyshev functions \psi(x) and \vartheta(x) . Let \Lambda(m) be the von Mangoldt function. Chebyshev functions are defined as \psi(x)\coloneqq\sum_{m\leq x}\Lambda(m),\quad\vartheta(x)\coloneqq\sum_{\begin{subarray}{c}p:\textup{prime}\\ p\leq x\end{subarray}}\log p. We denote non-trivial zeros of the Riemann zeta function as \rho=\beta+i\gamma with the real part \beta and the imaginary part \gamma . Let N(\alpha,T) be the number of non-trivial zeros of the Riemann zeta function in the range of \alpha\leq\beta\leq 1 and |\gamma|\leq T . Lemma 2.6 (Suzuki [ 9 ] ) . Assume that real numbers X,T satisfy 2\leq T\leq 2X . Then for 0\leq x\leq X , we have (5) \psi(x)=x-\sum_{\begin{subarray}{c}\rho\\ |\gamma|\leq T\end{subarray}}\frac{x^{\rho}}{\rho}+O(XT^{-1}(\log X)^{2}). Lemma 2.7 ( [ 5 , Theorem 6.1, p. 143] ) . (The Korobov–Vinogradov zero-free region) For s=\sigma+it , \tau=|t|+4 , \sigma>1-c_{0}(\log\tau)^{-\frac{2}{3}}(\log\log\tau)^{-\frac{1}{3}} , we have \zeta(s)\neq 0 , where c_{0}>0 is some absolute constant. Lemma 2.8 (Ingham [ 4 ] and Huxley [ 3 ] , [ 5 , p. 275] ) . For real numbers \alpha,T with \frac{1}{2}\leq\alpha\leq 1 and T\geq 2 , we have N(\alpha,T)\ll T^{c(\alpha)}(\log T)^{A},\quad c(\alpha)\coloneqq\left\{\,\begin{aligned} &\frac{3(1-\alpha)}{3\alpha-1}&(\frac{3}{4}\leq\alpha\leq 1),\\ &\frac{3(1-\alpha)}{2-\alpha}&(\frac{1}{2}\leq\alpha\leq\frac{3}{4}),\end{aligned}\right. where A is a constant. Remark 2.9 . Recently this result has improved by L. Guth and J. Maynard [ 1 ] . Using their results does not affect our main theorem for now. In the following, we evaluate several sums over non-trivial zeros of the Riemann zeta function. Lemma 2.10 (Suzuki [ 9 ] ) . For real numbers K,Y with 1\leq K\leq Y\leq X^{2} , we have \sum_{\begin{subarray}{c}\rho\\ K\leq|\gamma|\leq 2K\end{subarray}}Y^{\beta}\ll(Y^{\phi(\lambda)}+Y^{1-\eta+2\eta\lambda})(\log X)^{A}, where (6) \displaystyle\phi(\lambda)\coloneqq\left\{\,\begin{aligned} &\frac{3}{5}\lambda+\frac{3}{4}&(0\leq\lambda\leq\frac{25}{48}),\\ &3\lambda+2(1-\sqrt{3\lambda})&(\frac{25}{48}\leq\lambda\leq\frac{3}{4}),\\ &\lambda+\frac{1}{2}&(\frac{3}{4}\leq\lambda\leq 1),\end{aligned}\right. \displaystyle\eta=c_{1}(\log X)^{-\frac{2}{3}}(\log\log X)^{-\frac{1}{3}},\quad\lambda=\frac{\log K}{\log Y}, and constants A,c_{1}>0 . Lemma 2.11 (Suzuki [ 9 ] ) . For a real number \lambda with 0\leq\lambda\leq 1 , let \phi(\lambda) be the function given by ( 6 ), Then we have (7) \frac{3}{5}\leq\phi^{\prime}(\lambda)\leq 1. Lemma 2.12 (Suzuki [ 9 ] ) . Let \phi(\lambda) be the function given by ( 6 ) and consider following equations: \phi(\lambda_{1})-\frac{1}{2}\lambda_{1}=1,\quad\phi(\lambda_{2})+\frac{1}{2}\lambda_{2}=\frac{3}{2}. The